Pipeline Multitype Artifact Recognition Method Based on Inception_Resnet _V2 Structure Improving SSD Network

Table 1

Comparison of evaluation parameters of three target detection methods.

Models	Backbone network	mAP (%)	Detection accuracy (%)
			Equal elbow	Equal tee	Plug	Union	Cap
SSD	VGG-16	73.81	78.15	74.41	71.32	73.59	71.58
RCNN	VGG-16	77.74	72.18	79.68	75.72	78.33	70.21
DSSD	ResNET	78.27	82.08	80.05	73.84	77.14	78.22
DSOD	DenseNet	79.13	80.83	80.32	74.60	74.75	76.86
R-SSD	ResNet	80.04	82.35	81.67	73.62	75.83	80.02
Paper method	VGG-16	83.50	83.68	87.20	84.15	82.14	80.31